An Antisense Oligonucleotide Targeted to Human Ku86 Messenger RNA Sensitizes M059K Malignant Glioma Cells to Ionizing Radiation, Bleomycin, and Etoposide but not DNA Cross-Linking Agents¹

Ku86 plays an important role in a variety of cellular responses, either as a part of the DNA-dependent protein kinase multi-protein complex (DNA-PK)³ or as a component of the Ku heterodimer, and/or possibly on its own (reviewed in Refs. 1, 2). Ku86 and Ku70, at an equimolar ratio, form the Ku protein, an abundant nuclear protein originally identified as an autoantigen in patients with rheumatoid disorders. In vitro, Ku avidly binds ends of double-stranded DNA and is responsible for the DNA-binding activity of DNA-PK (3, 4, 5, 6, 7, 8). Inactivating mutations of either Ku86 or Ku70 genes result in a drastic decrease in the protein levels of both subunits, suggesting that one subunit of the Ku protein is required to stabilize the other (9, 10). The core protein of the DNA-PK multi-protein complex is represented by a M_r ∼460,000 catalytic subunit (DNA-PKcs) which, in itself, is a serine-threonine kinase similar to proteins of the phosphatidylinositol 3-kinase family (11). In vitro, DNA-PK is activated by free DNA ends (12, 13), is capable of phosphorylating several protein substrates, including the Ku heterodimer, and is also capable of autophosphorylation (reviewed in Ref. 14). Although the precise function of DNA-PK in vivo is not well defined, its importance in NHEJ and V(D)J recombination events has been shown clearly (reviewed in Refs. 15, 16, 17). Both NHEJ and V(D)J recombination involve the rejoining of DNA DSBs, the most cytotoxic form of DNA lesions in mammalian cells (18, 19). Indications that NHEJ and V(D)J recombination share the same machinery originated from work in C.B-17 scid mice that lack mature B and T cells as a result of a defect in V(D)J recombination (20). These mice were also found to be hypersensitive to IR attributable to a defect in DNA DSB repair (21, 22). Although single-strand breaks are the predominant lesions inflicted by IR and X-rays, it is the amount of DSBs in cellular DNA that correlates with cell death (23, 24). Experiments screening for defects in DSB repair and V(D)J recombination in X-ray-sensitive rodent cell lines identified three XRCC groups (25, 26). Assaying extracts from several XRCC groups for DEB activity, Rathmell and Chu (27) showed that three independently derived cell lines from the XRCC5 group were deficient in DEB activity. Later, it was shown that the DEB factor, which was absent in the XRCC5 group, was identical to the Ku86 gene product (28, 29, 30, 31, 32). Screening experiments in Ku86-defective xrs5 and xrs6 cells have also shown an increased sensitivity to DNA cross-linking agents such as nitrogen mustard, melphalan, and cisplatin (33, 34). These results are noteworthy in that Ku86 was shown to be involved in the repair of lesions (DNA cross-links), which are typically repaired by Ku86-independent excision repair pathways. Interestingly, when protein extracts isolated from B lymphocytes from untreated and chlorambucil-resistant CLL patients were assayed for DNA-PK activity, a linear correlation between the level of DNA-PK activity and the inherent in vitro sensitivity of CLL lymphocytes to CLB was found (35). It was also shown that increased DNA-PK activity was associated with increased DEB by the Ku heterodimer. Moreover, when wortmannin (a drug commonly used to modulate DNA-PK activity) was used at nontoxic concentrations, synergistic sensitization of B-CLL lymphocytes to the effect of CLB was achieved (36). Studies of a cisplatin-resistant and IR cross-resistant leukemic cell line (L1210) have also revealed an increased DEB activity associated with an overexpression of the Ku86 subunit, further indicating direct involvement of this protein in a resistant phenotype (37). However, some studies failed to show any correlation between DNA-PK activity and tumor cell radiosensitivity, or any differential expression of Ku86 in normal versus malignant cells (38, 39, 40).

To evaluate the role of Ku86 in cellular responses to treatment with different anticancer agents, we specifically inhibited its expression in M059K and M059J glioma cell lines by using highly specific 2′-O-methoxyethyl/uniform phosphorothioate chimeric ASOs directed against Ku86 mRNA. M059K and M059J cells, although isolated from the same tumor specimen, differ in their intrinsic sensitivities to IR and chemotherapeutic agents because of the absence of DNA-PKcs in the M059J line (41, 42). Having the potential advantage of specific binding to mRNA species and triggering their RNase H-directed cleavage, antisense ASOs have shown promise as therapeutic agents and research tools (reviewed in Refs. 43, 44, 45). Although most in vitro studies published thus far suggest a limited role for Ku86, such that it merely acts in partnership with Ku70 as a binding subunit for DNA-PKcs, its in vivo role is likely much broader. Therefore, using an antisense approach as a temporal knockout, the in vivo functions of DNA-PKcs and Ku86 can be discerned in different biological backgrounds, thus allowing for a more specific delineation of the biological roles of Ku86. Ultimately, Ku86 antisense ASOs have the potential of being used as therapeutic sensitizing agents used in conjunction with conventional therapies in the treatment of human malignancies. To this end, the following study will describe the development of a potent antisense inhibitor capable of specific down-regulation of cellular levels of Ku86.

Human EBV(−) WSU-CLL cells (46) were kindly provided by Dr. Ramzi M. Mohammad (Wayne State University School of Medicine, Detroit, MI) and were maintained in RPMI 1640 (Life Technologies, Inc., Burlington, Ontario, Canada) supplemented with 10% (v/v) FBS (Life Technologies, Inc.). M059K and M059J cells were obtained from American Type Culture Collection (Rockville, MD) and cultured in DMEM (Life Technologies, Inc.) supplemented with 10% (v/v) FBS. All cell lines were maintained in a humidified atmosphere of 5% CO₂ at 37°C.

To determine the most potent nucleotide sequence, 25 different 2′-O-methoxyethyl/uniform phosphorothioate chimeric ASOs, targeted to different sites of human Ku86 mRNA, were synthesized and kindly provided by ISIS Pharmaceuticals, Inc. (Carlsbad, CA). Antisense screening was performed on nonadherent WSU-CLL cells by electroporation. Briefly, cells were resuspended in antibiotic-free RPMI 1640 supplemented with 10% FBS. Cell aliquots were then mixed with different antisense oligonucleotides to achieve 2.0 × 10⁷ cells/ml and 10 μm ASOs in a total volume of 400 μl. After a brief incubation, aliquots of the cell/antisense combination were transferred into 0.2-cm sterile cuvettes (Bio-Rad, Hercules, CA) and electroporated with an Electro Cell Manipulator ECM 830 (BTX, a Division of Genetronics, San Diego, CA) using square pulses of 200 V. Postelectroporated cells were transferred onto individual p100 tissue culture dishes containing 10 ml of complete RPMI 1640 and incubated at 37°C, 5% CO₂ for 24 h.

Adherent M059K and M059J cells, when 70–80% confluent, were washed once with PBS and exposed to different concentrations of antisense and mismatch oligonucleotides combined with a Lipofectin solution (Life Technologies, Inc.) at a final concentration of 3 μg/ml in OPTI-MEM I medium (Life Technologies, Inc.) for 6 h. After transfection, cells were washed once with PBS to remove the Lipofectin solution and incubated in complete DMEM for various time intervals. For clonogenic assays, an additional 6-h transfection was performed 72 h after the initiation of the first transfection.

Total RNA was extracted using the RNeasy Mini kit (Qiagen, Santa Clarita, CA) according to the manufacturer’s methods. Extracted total RNA was treated with DNase enzyme in a DNA digestion mixture [10 μl of 100 mm MgCl₂, 10 mm DTT 1 μl of RNasin (Promega, Madison, WI), and 1 μl of RNase-free DNase 20 units/μl (Promega)] for 15 min at 37°C. After DNase treatment, samples were extracted once with phenol-chloroform-isoamyl alcohol and chloroform-isoamyl alcohol and precipitated with ethanol at −70°C. RNA concentration and purity were determined by measuring absorbance at 260 and 280 nm. The integrity of the samples was assessed by running 1 μg of RNA on a nondenaturing 1% agarose gel.

To quantitate Ku86 and Ku70 mRNA expression in antisense-treated and control cells, 1 μg of DNase-treated total RNA was used for first-strand cDNA synthesis, as described previously (47). The Ku86 PCR primers were 5′-CAG CCG ATT CAG CAA AAG TC-3′ (forward) and 5′-TCA TCC AAC CCA TCT TCA CC-3′ (reverse), with a predicted product size of 250 nucleotides. The Ku70 PCR primers were: 5′-TGA AGT GCT GTG GGT CTG TG-3′ (forward) and 5′-GGT CTC TTT GGC GTG AAC CT-3′ (reverse), with a predicted product size of 322 nucleotides. H3.3 histone was used as a housekeeping gene (48), and PCR primers were 5′-GTA AAG CAC CCA GGA AGC AA-3′ (forward) and 5′-ACG CTG GAA GGG AAG TTT G-3′ (reverse), with a predicted product size of 170 nucleotides. PCR was carried out in a total volume of 50 μl containing 5 μl of 10× PCR Buffer [500 mm KCl, 15 mm MgCl₂, and 100 mm Tris-HCl, pH 9.0 (Amersham Pharmacia Biotech)], 2.5 μl of deoxynucleoside triphosphates (2.5 mm each of dATP, dCTP, dGTP, and dTTP), 2 pmol of each forward and reverse Ku86 and H3.3 primer, and 1 unit of Taq polymerase (Amersham Pharmacia Biotech, Piscataway, NJ). PCR conditions were as follows: 1 cycle at 30 s/94°C, 29 cycles at 30 s/94°C, 60 s/55°C, 90 s/72°C, and 1 cycle at 5 min/72°C.

Whole-cell extracts were prepared by a single freeze-thaw cycle, as described (38). Briefly, cells were harvested, washed with ice-cold PBS and LSB [25 mm KCl, 10 mm NaCl, 1 mm MgCl2, 10 mm HEPES-KOH (pH 7.2), 0.1 mm EDTA, 2 μg/ml leupeptin (Sigma Chemical Co., St. Louis, MO), 2 μg/ml aprotinin (Sigma), 0.5 mm DTT, 0.5 mm phenylmethylsulfonyl fluoride], repelleted, and resuspended in LSB to a final volume of 2.5 times the original packed cell volume. Resuspended cells were incubated on ice for 10 min, frozen by plunging into liquid nitrogen, quickly placed at 37°C until slightly thawed, and then placed on ice. NaCl and MgCl₂ concentrations of fully thawed samples were adjusted to 0.5 and 10 mm, respectively, by adding HSB (5 M NaCl, 100 mm MgCl₂, and 5 mm DTT). Cell extracts were incubated on ice for 5 min and microcentrifuged at 10, 000 × g for 5 min at 4°C. Supernatants were collected, and protein concentrations were quantitated using the method of Bradford (49).

Cell extracts containing 10–50 μg of protein were separated by SDS-PAGE and transferred electrophoretically onto nitrocellulose membranes (Millipore, Bedford, MA) for 1 h in 25 mm Tris-HCl, 192 mm glycine, and 20% methanol. Membranes were blocked in TBS (Tris-HCl) containing 5% nonfat milk and probed with anti-Ku70 (N3H10), anti-Ku86 (111), and α-tubulin monoclonal antibodies (NeoMarkers, CA). Antigen/antibody complexes were detected by horseradish peroxidase-conjugated antimouse antibody and enhanced chemiluminescence (Amersham Pharmacia Biotech), according to the manufacturer’s instructions. The bands were quantified by densitometric analysis using Scion Image software (Scion Corp., Frederick, MD).

DNA-PK activity in M059K cells treated with Ku86 antisense was measured using SignaTECT DNA-PK assay system (Promega Corp., Madison, WI) according to the manufacturer’s methods. Briefly, after 72 h incubation, cells were lysed, and extracts containing 5–50 μg of protein were incubated with biotinylated peptide substrate supplied in the kit in the presence of Redivue [γ-³²P]ATP (Amersham) for 15 min at 30°C. Ten μl of each terminated reaction were spotted onto SAM^2TM membrane, washed, and dried. The amount of ³²P incorporated into the DNA-PK Biotinylated Peptide Substrate was determined by liquid scintillation counting.

For treatment with IR and cytotoxic anticancer agents, M059K and M059J cell lines were double-transfected with Ku86 antisense (ISIS101209) or its mismatch control (ISIS111449), as described above. At the end of the second transfection, cells were trypsinized and replated onto 6-well culture plates (Falcon; Becton Dickinson Labware, Lincoln Park, NJ) or 25-cm² tissue culture flasks (Sarstedt, Newton, NC) at various concentrations (200–15,000 cells/well). Cells on tissue culture flasks were exposed to γ-irradiation from a ¹³⁷Cs source for time periods equivalent to 0.5, 1.0, 1.5, 2.0, 2.5, 4.0, 6.0, and 8.0 Gy. For treatment with DNA-damaging agents, cells were allowed to attach to the bottom of the wells, washed with serum-free medium, and subsequently exposed to various concentrations of bleomycin (Bristol Laboratories), cisplatin (Sigma), etoposide (VePesid; Bristol Laboratories), and chlorambucil (Sigma). After 14 days, cells were washed with PBS, fixed with gluteraldehyde, and stained with 1% crystal violet. Washed dishes were dried, and colonies containing >50 cells were scored and plotted on a log scale against drug concentration as a fraction of control.

To select the most potent site for antisense targeting, a panel of 25 oligonucleotides (predicted to hybridize to different regions of the human Ku86 subunit of DNA-PK; Complete cDNA, NCBI Accession J04977) were synthesized using a mRNA walking approach (50). WSU-CLL cells were transfected, by electroporation, with 10 μm oligonucleotides, and the levels of Ku86 mRNA were determined by semiquantitative RT-PCR after a 24-h incubation as described in “Materials and Methods.” A reduction in Ku86 mRNA level was observed upon treatment with several ASOs (Fig. 1). However, the degree of down-regulation varied greatly. Some ASOs had little or no effect. Only two ASOs, designated as ISIS101209 and ISIS101212, were capable of down-regulating Ku86 mRNA levels by 90% of the initial steady-state level and thus were used for subsequent experiments.

Because of drawbacks of electroporation experiments, such as limited transfection efficiency and high cell death, the specificity and dose-response of ISIS101209 and ISIS101212 ASOs were studied in M059K and M059J adherent cell lines using a cationic lipid transfection reagent, Lipofectin. These cells were transfected with Ku86 ASOs (ISIS101209 and ISIS101212) and their mismatch controls (ISIS111449 and ISIS111450, respectively) at concentrations ranging from 25 to 400 nm. However, treatment of both M059K and M059J cells with ISIS101212 was associated with a high level of cell detachment. Thus, subsequent experiments were conducted using ISIS101209 (5′-CCC ATG AAG AAT CTT CTC TG-3′) and its respective mismatch control, ISIS111449 (5′-CCT ATC AAT AAG CTC CTG TG-3′). After a 12-h incubation, a dose-dependent reduction of Ku86 steady-state mRNA levels was observed in both cell lines transfected with ISIS101209. Specifically, we observed a 50% effective concentration (EC₅₀) of <25 nm and mRNA levels <5% of controls when antisense was used at concentrations >200 nm (Fig. 2, A and D). Treatment with the mismatch control (ISIS111449) did not lead to alterations in Ku86 steady-state mRNA levels (Fig. 2, B and D). These results indicate that treatment with ISIS101209 ASO leads to a substantial decrease in Ku86 steady-state mRNA level, at low concentrations, and in a dose-dependent manner in both cell lines tested. The specificity of antisense effect was confirmed when mismatch control was used to transfect both cell lines, and no effect on Ku86 mRNA expression was observed. Similarly, mRNA steady-state levels of Ku70 binding component of DNA-PK complex that shares some level of sequence homology with Ku86 mRNA were not found to be altered by the treatment. When the same RNA extracts were assayed by RT-PCR, with the primers to human Ku70, steady-state levels of Ku70 mRNA were constant at all antisense concentrations tested (Fig. 2 C). This experiment also confirmed the high degree of specificity of ISIS101209 antisense for its target, Ku86.

To investigate the effect of a single Ku86 antisense treatment on Ku86 mRNA expression over a given time period, M059K and M059J cells were transfected with 200 nm ISIS101209 and ISIS111449 mismatch control, for 6 h, in the presence of 3 μg/ml Lipofectin in OPTI-MEM I. When the transfection medium was replaced with DMEM complete medium, the cells were incubated so that the time from the beginning of transfection corresponded to 12, 24, 48, 72, and 96 h. After these incubations, the cells were lysed, and Ku86 mRNA levels were measured by RT-PCR analysis, as described in “Materials and Methods.” Transfection with Ku86 antisense ISIS101209 led to a rapid decrease in steady-state levels of Ku86 mRNA (<5% of ISIS111449 mismatch-treated control) and remained at this level for ∼72 h. However, the level increased to ∼20% of the control values by 96 h (Fig. 3). As was expected, no effect on Ku86 mRNA levels was detected in cells transfected with mismatch-control. It is not likely that the observed decrease in Ku86 steady-state mRNA level was attributable to a generalized cell kill, because the same extracts, when assayed for the levels of the housekeeping gene and Ku70 mRNA expressions, showed the levels similar to the untreated controls. These results demonstrate that a single transfection of both M059K and M059J cell lines efficiently suppressed steady-state Ku86 mRNA levels for ∼3 days.

It was reported previously that the apparent half-life of Ku86 protein in the K562 human erythroleukemia cell line was >5 days (51). Because a single transfection of M059K and M059J cells resulted in a substantial elimination of Ku86 mRNA levels for slightly more than 72 h, additional transfections would be required to study the effect of Ku86 antisense treatment on Ku86 protein suppression and related phenotypic changes. However, in a time course experiment that followed, a single antisense transfection showed that Ku86 protein levels (detected by Western analysis) decreased to ∼50% of control levels by 48 h and to <10% of control levels by 72 h (Fig. 4,A). By 96 h, Ku86 protein levels were equal to control levels, although Ku86 mRNA levels represented 20% of control levels. In contrast, treatment of M059K and M059J cells with ISIS111449 mismatch control did not affect Ku86 or Ku70 protein levels (Fig. 4 B).

Previous experimental findings indicated that the absence of Ku86 protein in xrs6 cells renders Ku70 protein levels undetectable despite normal expression of the Ku70 mRNA, suggesting the importance of the Ku86 gene product for the stability of the Ku70 protein (9, 10). To examine whether a decrease in Ku86 protein levels in M059K and M059J cells treated with Ku86 antisense would affect the expression of the Ku70 subunit, we probed the same membranes used for Ku86 protein analysis with Ku70-specific antibodies. We observed a marked decrease in Ku70 protein levels associated with low levels of Ku86 expression 72 h after transfection of both cell lines with Ku86 antisense (Fig. 4). Mismatch controls showed no effect on Ku70 or Ku86 levels. These findings are in agreement with previous reports in that each subunit of the Ku protein is required to stabilize the other (9, 10, 52).

Upon binding to DNA lesions, such as DNA DSBs, DNA-PK catalytic subunit exhibits serine/threonine kinase activity toward many proteins, thus its role in DNA damage signaling can be well suited. Therefore, next we decided to examine whether Ku86 antisense treatment of M059K cells proficient in DNA-PK catalytic subunit could abolish the kinase activity of the enzyme. After 72 h incubation, cells were lysed, and extracts containing 5–50 μg of protein were assessed in the presence of DNA-PK biotinylated peptide substrate and [γ-³²P]ATP with or without calf thymus double-stranded DNA for 15 min at 30°C. It was found that DNA-PK catalytic activity was significantly diminished in the cells treated with Ku86 ASOs, and a 5-fold difference was observed when whole-cell extracts containing 50 μg of protein from Ku86 antisense- and mismatch-treated M059K cells were compared with each other (Fig. 5).

Colony survival assays were used to investigate the effect of Ku86 protein down-regulation associated with antisense treatment on the sensitivity of M059K and M059J cells to IR and various DNA-damaging agents. Both cell lines were transfected two times for 6 h with 200 nm antisense or mismatch control within a 72-h interval. At first, we determined whether treatment of cells with Ku86 antisense and its mismatch control would affect cell sensitivity to IR. Treatment of M059K and M059J cells with mismatch control did not confer any additional sensitivity to IR (ID₉₀; 4.3 Gy and 1.26 Gy, respectively) and the survival curves (Fig. 6, A and B) were similar to curves published previously for these cells (41). However, treatment of M059K (DNA-PKcs-positive) cells with 200 nm Ku86 antisense was associated with a 2-fold potentiation of radiation sensitivity (Fig. 6, A and B) compared with mismatch-treated controls (ID₉₀; 2.0 Gy versus 4.3 Gy; P < 0.001, paired t test). No effect was observed in M059J (DNA-PKcs-deficient) cells after Ku86 antisense treatment. These findings indicate that treatment of M059K cells with Ku86 antisense enhances the radiosensitivity of these cells to levels similar to their M059J DNA-PKcs mutant counterparts. Similarly, when cells from the same transfection experiment were assayed for their sensitivity to bleomycin and etoposide, it was found that Ku86 antisense-treated M059K cells also became more sensitive to the drugs when compared with mismatch-treated controls (ID₉₀; 10.7 mu/ml versus 21.0 mu/ml and 12 μm versus 22.9 μm, respectively; P < 0.001, paired t test; Fig. 6, C and D). Interestingly, although Ku86-antisense treated M059K and M059J cells exhibited mismatch-treated control levels of sensitivity to cisplatin and chlorambucil, Ku86 antisense treatment of M059J cells led to a 2-fold potentiation of the effect of etoposide (Fig. 6,E). The results from IR and DNA-damaging agent treatment experiments are summarized in Table 1.

Accumulating evidence suggests the implication of the Ku heterodimer in tumorigenesis and acquired drug resistance (35, 36, 53, 54, 55, 56, 57, 58). As such, we have developed highly specific 2′-O-methoxyethyl/uniform phosphorothioate chimeric ASOs directed against Ku86 mRNA, and using antisense strategy, we have markedly suppressed the expression of both Ku86 mRNA and protein in two human glioma cell lines (M059K and M059J). Because of a defect in a key protein of the NHEJ DNA repair pathway in M059J cells, these two cell lines differ in their inherent sensitivity to IR and chemotherapeutic agents. Cell lines mutated in the different components of the DNA repair pathways represent potentially useful models for elucidating mechanistic aspects of the repair process. Despite the existence of a variety of rodent cell lines containing mutations in proteins of the NHEJ repair pathway and thus defective in DNA DSB repair, only one glioma-derived cell line of human origin (M059J) has been described thus far (41). M059J cells are hypersensitive to ionizing radiation and, because of a frameshift mutation in the PRKDC gene, fail to express protein for the catalytic subunit of DNA-PK (42, 59). The M059K cell line is isogenous to M059J and was established from a different area of the same tumor. It exhibits a radioresistant phenotype and expresses functional DNA-PKcs protein. Because both cell lines express the Ku heterodimer, we investigated whether transient knockout of Ku86 would further increase IR sensitivity in DNA-PKcs-deficient cells (M059J) and whether depletion of Ku86 in DNA-PKcs-proficient M059K cells would increase their IR sensitivity to levels similar to those of M059J cells. As expected, inhibition of Ku86 gene expression by Ku86 ASOs led to an enhanced radiosensitivity in DNA-PKcs-proficient M059K cells but not in DNA-PKcs-deficient M059J cells. Similarly, inhibition of Ku86 gene expression by Ku86 antisense ASOs led to an increased sensitivity to bleomycin only in DNA-PKcs-proficient M059K cells but not in DNA-PKcs-deficient M059J cells. However, Ku86 antisense treatment did not render M059K cells as sensitive to IR and bleomycin as DNA-PKcs-deficient M059J cells. Interestingly, when both antisense-transfected M059K and M059J cells were treated with etoposide, a topoisomerase II inhibitor that generates strand breaks in DNA (60, 61), enhanced sensitivity was observed in both cell lines, and the level of sensitivity of antisense-transfected M059K cells was similar to the level of DNA-PKcs-deficient M059J cells. It was reported previously by Jin et al. (62) that Ku-deficient cell lines, but not cell lines defective in DNA-PKcs, were hypersensitive to etoposide, implying that etoposide-induced DNA repair is DNA-PK holoenzyme independent. However, it was found that wortmannin, a drug commonly used to modulate DNA-PK activity, potentiated the cytotoxicity of etoposide in the Ku-proficient CHO-K1 cell line but not in the Ku-deficient xrs-6 cells (63). Similarly, findings by Johnson and Jones (64) also demonstrated that the hamster XRCC7 cell line deficient in DNA-PKcs was 3-fold more sensitive than the DNA-PKcs-proficient V79 cell line. Therefore, these studies suggested that DNA-PKcs might be responsible for the etoposide-resistant phenotype. In the present work, we also showed the implication for DNA-PKcs in this repair process. Moreover, greater sensitization to etoposide was found when neither Ku86 nor DNA-PKcs proteins were expressed, implying that two independent DNA repair pathways may operate in etoposide-induced DNA damage response, and that the two proteins are needed for efficient DNA damage repair. We also speculate that the Ku-directed mechanism responsible for the etoposide damage repair may be DNA-PKcs independent, whereas a DNA-PKcs-directed mechanism requires the presence of the functional Ku component. Although it is generally believed that DNA-PKcs activity is dependent on its interaction with the Ku component, Ku and DNA-PKcs mutations may still present dissimilar phenotypic manifestations. One possible explanation for these differences emanates from work in mice deficient in various components of the NHEJ repair pathway. Although mice deficient for DNA-PKcs exhibit compromised immunity and increased radiosensitivity, the absence of Ku confers additional defects such as atrophic skin and hair follicles, osteopenia, premature growth plate closure, hepatocellular degeneration, and age-specific mortality (65). Detailed analysis of DNA-PKcs-deficient cells derived from scid mice revealed that, although they are able to produce signal joints, they are unable to produce coding joints during V(D)J recombination events. The cells derived from Ku86^−/− animals, on the other hand, are unable to generate both signal and coding joints (for details, see Ref. 66). Given that the murine scid mutation results in the truncation of only a small portion (2%) of the DNA-PKcs COOH terminus, it has been argued that the protein (DNA-PKcs) may still retain partial functionality sufficient to assist in signal joint formation. However, when signal and coding joint formation was examined in M059K and M059J cells, it was found that M059J cells, despite lacking any detectable amounts of DNA-PKcs, expressed signal joints at levels similar to those of wild-type (67). These data suggest that Ku86^−/− and DNA-PKcs^−/− mutants differ in their ability to resolve DNA DSBs. Another line of evidence for distinct functions of Ku and DNA-PKcs proteins comes from IR-induced replication arrest experiments. When replication arrest was induced in DNA-PKcs-proficient cells (M059K and HeLa) and Ku86-deficient cells (xrs5), normal kinetics of recovery were observed, whereas no recovery was observed in DNA-PKcs-deficient cells (irs20 and M059J; Ref. 68). Because both xrs5 and irs20 cell lines share DNA DSB joining defects, it might be concluded that recovery of DNA replication may be independent of both Ku and efficient DNA DSB rejoining and may simply require the presence of a functional catalytic subunit (DNA-PKcs).

The contribution of DNA-PK to DNA cross-linking and alkylating drug resistance has been somewhat ambiguous. In addition to their higher sensitivity to IR, M059J cells were also found to be more sensitive to treatment with nitrogen mustard and 1,3-bis(2-chloroethyl)-1-nitrosurea (41). Similarly, Caldecott and Jeggo (33) have shown increased sensitivities to nitrogen mustard, melphalan, and cisplatin in rodent Ku mutant cells. Nitrogen mustard, chlorambucil, and melphalan represent classical bifunctional alkylating agents, which form DNA interstrand and intrastrand cross-links, as well as monofunctional DNA adducts (69, 70). Similarly, cisplatin, in addition to inducing the formation of monofunctional DNA adducts, DNA interstrand and intrastrand cross-links can also cause DNA-protein adducts (71, 72). Of these lesions, DNA intrastrand cross-links are considered to be the most cytotoxic. When several cell lines were treated with DNA cross-linking agents and subjected to pulsed-field gel electrophoresis, it was found that DNA DSBs were rapidly generated. Unlike IR or bleomycin, DNA cross-linking agents do not directly produce breaks in DNA strands. Although the mechanism by which DNA intrastrand cross-links are repaired in mammalian cells is not clear, it is commonly believed that DNA DSBs are generated as intermediary products during DNA repair (33, 73, 74). Therefore, the presence of functional components of the NHEJ repair pathway, the predominant pathway for DNA DSB repair in mammalian cells, would be critical for efficient break rejoining. In agreement with this hypothesis, previous work in our laboratory showed a linear correlation between the level of DNA-PK activity and the inherent in vitro sensitivity of CLL lymphocytes isolated from previously untreated (sensitive) and chlorambucil-treated (resistant) patients (35, 36). In contrast, our present results indicate that depletion of Ku86, using an antisense approach, did not result in any additional sensitivity to cisplatin or chlorambucil in M059K DNA-PKcs-proficient cells. One possible explanation would be that repair of the damage inflicted to DNA by cisplatin and chlorambucil, at least in the cell lines studied, does not generate a sufficient number of double-strand intermediary products, inefficient repair of which in a Ku86-deficient environment leads to cell death. Unlike G₀-arrested B-CLL lymphocytes that exhibit 2N ploidy, both M059K and M059J cells rapidly proliferate without showing G₁ checkpoint arrest (60, 75). It is possible that when these cells reach late S and G₂ with 4N ploidy, homologous recombination becomes the predominant mechanism responsible for efficient removal of the damage induced by DNA cross-linking agents. Nevertheless, one may argue that a leaky phenotype, attributable to some residual amounts of Ku86 protein in Ku86 antisense-treated cells, is responsible for the NHEJ-dependent DSB rejoining. However, this argument does not explain our observations of a difference between the response of Ku86 antisense- and mismatch-treated M059K cells to low doses of IR and bleomycin. On the other hand, we cannot claim that the nature of the breaks produced by IR is identical to the breaks produced during enzymatic removal of the DNA cross-links, and as such, cannot assume that the same repair proteins are involved in their rejoining. If DNA DSBs produced by DNA excision mechanisms can be directly sealed by DNA ligases, then direct ligation of IR-induced strand breaks is not likely to occur (23). Specifically, this scenario is unlikely because the direct rejoining of DNA by DNA ligases requires closely apposed 5′-phosphate and 3′-hydroxyl groups, whereas IR produces a nucleotide “gap” that separates 5′-phosphate groups from 3′ termini (61). Overall, it can be concluded that the involvement of Ku and DNA-PKcs in DNA DSBs induced by DNA cross-linking agents may be different in G₀-arrested and rapidly proliferating cells. Another possible explanation for the differences observed in the glioma cell sensitivities may be related to intrinsic differences in genes other than DNA-PKcs. For example, fluorescence in situ hybridization analysis of both M059J and M059K cell lines indicated that, when compared with M059K, M059J cells are near pentaploid with a much broader distribution of chromosome numbers (59). This finding indicates that, because of a greater level of abnormal chromosomal segregation and genomic instability, M059J cells have accumulated larger amounts of genetic abnormalities than have M059K cells, which ultimately result in decreased cell survival. Indeed, the first report on M059K and M059J cells showed that the in vitro plating efficiency of M059J cells was almost two times less than that of M059K cells (0.10 and 0.23, respectively; Ref. 41).

In the context of earlier work, it is noteworthy that 72 h after a single in vitro transfection of two human glioma cell lines with 200 nm Ku86 ASOs, a dramatic decrease in Ku86 mRNA levels, accompanied by a decrease in Ku86 protein levels, were observed. A previous report on Ku86 protein turnover in a human erythroleukemia cell line (K562) showed an apparent half-life of >5 days (51). Similarly, human epidermal basal cells that do not synthesize this protein have been shown to contain a significant amount of Ku (74). In contrast, analysis of peripheral blood neutrophils, the postmitotic life span of which is ∼6.5 days, revealed an absence of Ku (51). Although the issue of Ku86 protein turnover under antisense treatment was not fully addressed in this work, we speculate that this phenomenon of rapid Ku86 disappearance is related to the activation of pathways responsible for the degradation of active Ku86 protein. To exclude the notion that the glioma cell lines used in our study were somehow responsible for the accelerated turnover, we transfected MCF7 and T98G cells with Ku86 antisense and found similar kinetics of Ku86 protein degradation.⁴ Previous experimental findings indicated that the absence of Ku86 protein in xrs6 cells renders Ku70 protein levels undetectable despite normal expression of the Ku70 mRNA, suggesting the importance of the Ku86 gene product for the stability of the Ku70 protein (9, 10). Li et al. (52) has also found that heterozygous Ku86^+/− human HCT116 cells contained 20–50% as much Ku86 and Ku70 protein as the parental cells. Similarly, in the present work, when the same membranes were probed with the antibodies against Ku70, we observed a substantial decrease in Ku70 protein level that was associated with low expression of Ku86, whereas Ku70 mRNA level remained unaffected. Interestingly, we observed that 96 h after a single Ku86 antisense transfection, Ku86 protein levels returned back to control values when Ku86 mRNA levels were only 20% of control. Although this issue was not fully addressed in the present work, it is likely that newly synthesized Ku86 protein by avid binding to Ku70 protein either becomes more stable or is subjected to slower protein turnover, suggesting that not only stability but also turnover of 1 Ku subunit is dependent on the other. The fact that both Ku subunits are functionally dependent on each other and that Ku70 and Ku80 exist in organisms ranging from yeast to humans strongly supports this hypothesis. However, these findings are in disagreement with a previous report demonstrating that transfection of SV40-transformed MRC5V1 human fibroblasts with a vector (pcDNA3) containing a Ku86-antisense cDNA caused these cells to exhibit decreased Ku86 protein expression and a radiosensitive phenotype, without a decrease in Ku70 expression (63).

In conclusion, our cross-resistance experiments failed to demonstrate an involvement of Ku86 in DNA cross-linking agent (chlorambucil and cisplatin) resistance in human glioma cells. Our results suggest that Ku86 antisense administration, combined with radiotherapy and/or radiomimetics (such as bleomycin), may serve as a useful treatment to overcome radioresistance in several malignancies. Lastly, we have demonstrated the viability of using Ku86 antisense as a means for temporal knockout of Ku86, thus encouraging future research in the elucidation of the precise role of Ku in the realm of DNA repair and other potential biological arenas.